I-AI Economy Inokufuneka I-Engine Enye: Kwi-Radical Vision ye-Ambient

Kule edition esizodwa yethuPhantsi kweStartupIshan Pandey wahlala kunye neTravis Good, co-founder ofUmgangatho, inethiwekhi ye-AI eyahlukileyo ebonakalisa indlela yokucubungula, ukulawulwa, kunye ne-blockchain consensus.

Ukwakhiwa kwezinhlelo zokuthutha ezifanelekileyo kumathemikhali ukuya ku-AI-driven biotech efficiency, uTravis ngoku uqhagamshelwano wayodecentralizing AIKule nqakraza elidlulileyo kunye nenkqubo, i-Travis ibonise ukuba i-Ambient ibonise iingcebiso ze-industry, ngoko i-Proof of Work ibonelela, kwaye indlela yokuba i-AI inokufunda i-internet economy entsha enokutholakalayo.

Ishan Pandey: Hi Travis, welcome to our Phantsi kweStartupUyaqala eHarvard, usebenza ekuphuculeni iinkqubo ezincinane kwi-biotech kunye ne-transport, kwaye ngoku uyenza inkampani ye-blockchain eyenziwe nge-AI. Uyakwazi ukuhamba nathi ngexesha elandelayo? Yintoni iye iye yenza i-shift yakho ukusuka kwizicelo ze-AI ezivamile kuquka inethiwekhi ye-AI eyahlukileyo?

Travis Good:Nceda uxhumane, Ishan. Ndingathanda kwiindawo ezinzima apho abaculi zibonisa ukuba iingxaki ezifanelekileyo ziquka "umgangatho olufanelekileyo." Oku kwakhona kwimeko ezintathu njengoko uye wabelane, i-biotech, i-transport, kwaye ngoku i-blockchain.

Kwi-biotech, ndingathanda kwi-reunion kunye ne-pharmaceutical company enkulu xa baye baye baye baye baye baye baye baye zibonise iindidi ezimbini yokuguqula i-pesticide eyodwa usebenzisa indlela epheleleyo ye-brutal force kunye neempawu ezininzi kunye nezixhobo ezisebenza 24/7. Yonke into ndingathanda ngexesha yaye ndingathanda ukuba "ke yinto ezininzi ezihlabathi." Kule inkampani apho ndiye i-CTO, sisiza ukunciphisa ngokugqithisileyo ukutya kwezinto ezininzi, usebenzisa i-machine learning kunye ne-computational chemistry. Abavela ekuqaleni ngenxa yeengxabiso kodwa ekugqithisileyo i-betting iye yahlulwe ngenxa yok

Ngokufanelekileyo, kwi-transport, iingcali wabelane ukuba iinjini ze-regulatory (okungabikho kwizigaranti ze-performance kodwa ingasebenza ngokukhawuleza), kunye ne-algorithms ze-genetic (okungabikho kwizigaranti ze-performance kunye ne-timing) ziyi-best ye-industry ye-freight ingathanda ukufumana. I-scepticism yam yaye iye yandise i-crab-hole ye-research eyenza ngexesha leyo kwaye ngexesha lokuphumelela i-speed ye-University of Rome njengezinxalenye abanamakhasimende ehlabathini yokuqala ye-matematically ye-optimal freight railway movement planner, eyenza ngokukhawuleza, yaba i-per

I-CryptoAI, ebonakalayo ukususela ngo-2017, iye yenza iingxaki ezininzi ezininzi ezininzi ezibonakalayo. Okokuqala, ukuba i-"marketplaces yeemodeli" ziyi-best for network economics, innovation, and performance. I-second is that Proof of Stake is the best way to secure networks that feature a lot of mining. Ndingathanda kakhulu kunye nabanye iingxaki, kodwa zithunyelwa kwakhona kwi-chain-by-chain. Emva koko, i-Centralized AI waya kwi-scene kunye ne-"OpenAI" kunye ne-Anthropic, kwaye ndiza ukuqala ukuxhaswa ngokwenene, ngenxa yokuba akukho umntu waza ukwakha into efanelekileyo ye-decentralized economic alternative. Ndingathanda is

Ishan Pandey:I-Ambienst iye yenza i-$7.2 million yeengxaki ezijoliswe yi-Crypto Startup Accelerator ye-a16z, kunye ne-Delphi Digital kunye ne-Amber Group. Abathengisi bafuna iimodeli ezininzi zebhizinisi kunye neengxaki ezininzi ezibonakalayo. Yintoni iingxaki zeemodeli zentengiso ze-Ambienst kunye nokufanelekileyo kwimeko ye-economic?

Travis Good:Ndingathanda ukuchitha iingcebiso kumntu, kodwa ndingathanda ukuba abathengisi bethu bayazi ukuba i-Ambient isixazululo kwinkcebiso esisiseko se-economic eyenza njengoko i-AI ibonelela kwi-silicon ye-economy yehlabathi.

imodeli yethu yentengiso lihlanganisa malunga neyintoni ethetha "ukubuyiselwa ukusebenza" - inkqubo apho ama-miners zithenga kunye ne-inflation kunye ne-transaction-based rewards ngokwenza i-AI inference ebonakalayo ukuba abasebenzisi kubunzima kwaye baye zithengise. Ngokungafani ne-cryptocurrencies ezivamile apho i-mining ayikwazi ukuvelisa i-utility ngqo, ama-ambient miners zibonisa i-value ye-economic nge-transaction nganye. Oku kwenza i-virtuous cycle: xa abasebenzisi abaninzi zinobunzima i-AI inference, ama-miners abaninzi zihlanganisa ukuba zithengise, ukwandisa inethiwekhi kwaye ukwandisa i-utility yayo.

I-defendability ifumaneka kwiintlobo ezintathu eziphambili. Okokuqala, i-innovations yethu zobuchwepheshe kwi-confirmed inference-ukufumana i-overhead ye-0.1% kuphela kunokuba i-10-1000x i-overhead iindleko zeendiza. Okokuqala, i-focus yethu kwi-standardization-ukuphumelela kwimodeli ye-high-quality ye-big-language ngaphandle kokuphumelela iimveliso kwiimodeli ezininzi, sincoma kakhulu imveliso ye-miner. Kwakhona, iimiphumo ze-network-ngokuqhelekanga iimveliso ze-miner, ukunciphisa i-latency kunye nokuphumelela kwimveliso, yenza i-network engaphezulu kubasebenzisi.

Ukusuka kuluhlu, i-Ambient ivela i-economy apho i-currency ifumaneka ngqo kwi-resource ye-economic engundoqo ye-era elandelayo: i-machine intelligence. Xa i-AI ibonakala kubalulekile ekubeni zonke iinkqubo zebhizinisi, i-Ambient ibonelela kunye ne-infrastructure ye-computing kunye ne-framework ye-financial ukunceda lo mveliso.

I-aspekti ebonakalayo ngoko ndingathanda ukuba lula ukubonisa, ngaphandle kokufunda kuphela, iimvavanyo ezininzi - umzekelo, ngexesha lokufunda iimvavanyo yethu, sisetyenziswa kwinkqubo yethu yokuqiniswa kwimodeli ukuya kwi-400 billion iiparamitha.

Ishan Pandey:Ukukhuthaza i-capital kwimarike yamanje kunzima, kodwa i-Ambient yenzelwe ngempumelelo ukusuka kwi-a16z, i-Delphi Ventures, kunye ne-Amber Group. Yintoni iingcebiso ezininzi ezifundweni kwinkqubo yokukhuthaza iingcebiso, kwaye iingcebiso efanelekileyo ze-AI kunye neengcebiso ze-blockchain ezinikezele ukufumana abathengisi abaphambili?

Travis Good:Ngokusho izifundo ze-fundraising, ndingathanda ukuba ndingathanda ukuba i-technical diligence engapheliyo ye-fond, i-impression kunye ne-excitement engaphezulu, ngenxa yokuba le nqakraza ibonisa i-technical capabilities ye-fond, iingcango kunye ne-will to engabandakanyeka kwi-project. Umzekelo, ndingathanda kakhulu ukuba i-a16z CSX ibonise i-panel epheleleyo yeengcali ezaziwayo ukuba ibonise i-Proof of Logits yethu ngokugqithisileyo.

I-Lesson elinye ndingathanda ukuba ayikho bonke unayo i-thesis yonke into, ngoko kubalulekile kwaye kufuneka uqhagamshelane neengxaki eziquka i-thesis kwi-area ye-projekt yakho kunye ne-thesis yaye ilungele kwi-project yakho. Siye kwenziwe kakuhle kwi-front efanayo kunye nabanye abathengisi bethu, ndingathanda, kwaye oku kuthetha elungileyo.

Ngokutsho iingcebiso zayo? Ndingathanda iintloko kubalulekile. I-Ambient isiseko kuyinto ukunikezela i-AI eyahlukileyo kwinqanaba, ukufumana iingcebiso zezimali ezisemgangathweni ze-AI eyahlukileyo, kwaye ibonelela i-miners iingcebiso ye-economic efanelekileyo ngokuvumela kunye nokuguqula i-Proof of Work. Thina ufuna ukunceda kuphela inethiwekhi yethu, kodwa ukunikezela i-Web2 kunye nezinye i-blockchains kunye neengcebiso ezininzi kunye neengcebiso ezininzi ezincinane ze-confirmed kwimodeli ye-open weights engcono. Ndingathanda kakhulu ukuba abathengisi bethu ziye zihlanganisa le nq

Ishan Pandey:Yintoni ufuna ukuhanjiswa kwimali entsha kwi-R&D, i-infrastructure, kunye neengxaki zokusebenza? Yintoni iindlela ezizodwa zenzulululwazi okanye iinkqubo zokusebenza kwimarike?

Travis Good:Thina usihlanganisa imali kwiintlobo ezintathu eziphambili, kunye nokuphuhlisa ubuchwepheshe kubandakanyeka kakhulu.

Okokuqala, malunga ne-60% kubandakanyeka kwi-R & D ukwandisa inkqubo yethu ye-inference ye-verified kunye nokwandisa i-fork yethu ye-Solana. Thina ufuna i-inference ye-verified ukuba yinkonzo efanelekileyo kwi-miners ezininzi kunye ezincinane. Ngokusho, nathi ukuba i-API yethu yinto efanelekileyo kumadivayisi, ngoko ke sisebenza kakuhle kwi-Solana ingxenye.

Okokuqala, malunga ne-25% yenzelwe kwi-infrastructure kunye nokuthuthukiswa kwe-testnet. Sinikezela ukulungiselela i-testnet ngexesha elide yonyaka elide, leyo iyavumela i-miners kunye ne-developer ukufumana i-platform kwi-first-hand. Oku kubandakanya ukwakhiwa kwezixhobo ze-developer, i-documentation kunye ne-SDK zokwenza ukuhlanganiswa ngempumelelo.

I-15% ezidlulileyo isekelwe kwi-go-to-market strategy yethu, ebandakanya iintlobo ezimbini: ukuhweba kwe-miner kunye neengxaki ze-developer. Kwi-mining side, sinikezela izixhobo kunye neengxaki zokunceda abathengi be-GPU ukuhambisa kwi-Ambient Mining. Kwi-developer, sinikezela ukubonisa iimeko ezininzi zokusetyenziswa kwe-inthanethi ye-AI. Ukuba unomdla, i-developer, okanye i-platform ye-developer (njenge-providers ye-agent framework) siyafuna ukufumana kwakhona!

Ukusetyenziswa oku kutshintsha iziphumo ezininzi ezininzi kuquka ukusuka kuTestnet yethu, ukufikelela kwinqanaba lethu yokuqala yama-miners, ukufikelela kwinqanaba lethu le-cross-chain for interoperability, kunye nokuthuthukiswa kwe-publish-facing API gateways zethu.

Ishan Pandey:I-AI inference kwi-scale yi-challenge emangalisayo, ikakhulukazi ekubaleni ukhuseleko, i-decentralization, kunye ne-cost-efficiency. Yintoni i-Architecture ye-Ambient ibonise i-compromise phakathi kwe-computational efficiency kunye ne-blockchain verification?

Travis Good:Kwicandelo lwe-Architecture ye-Ambiente i-proof of logits system, nto leyo ibonisa ukuchithwa kwimeko ye-verified inference enikezela iingxaki ze-traditional.

Iinkqubo ezininzi ze-verification ziquka i-binary choice: okanye i-offering efficiency ngenxa yobugcisa (njenge-zkSNARKs nge-1000x overhead) okanye i-offering security ngenxa yobugcisa (njenge-optimistic verification systems). I-innovation yethu ibonelela ukuba i-"fingerprint" ye-inthanethi ye-AI model - i-outputs ye-numerical ebizwa ngokuba yi-logits - ingasetyenziswa ukulayisha iimodeli nge-overhead engaphansi.

Yintoni iindlela zokusebenza: Xa imodeli ye-AI ikhiqiza umbhalo, inikeza i-logits eyenziwe ngalinye ngalinye ye-token. Lezi zibonelelo lwezimo zangaphakathi kwimodeli kwaye zinokufumaneka kwi-hash to create a compact representation. Iingcebiso esiyingcali yaba ukuba i-validators ayidinga ukuba zibonele inqubo ye-generation ngokupheleleyo - zibonele i-tokens ezahlukileyo kunye neengxaki ze-matematical phakathi kwabo ngexesha elifanelekileyo, ukunciphisa kakhulu iimfuno ze-computational nangokufumana iinkonzo zokhuseleko ezininzi.

I-system iye yenzelwe njenge-non-blocking proof-of-work consensus mechanism, nto leyo kuthetha ukuba ukulawulwa kwenziwa ngexesha elifanelekileyo kunye ne-transaction processing. Oku ukugcina i-high-throughput ye-Solana-based foundation yaye ukongezelela izinzuzo ezininzi ze-proof-of-work mining.

Ukusetyenziswa kwe-decentralization, sinikezela iindlela ze-sharding ze-models ezihlangene neziphumo ezidlulileyo ze-academic. Oku kuthetha ukuba sinikezela iimodeli ezininzi (600B + iiparametres) kwiiyunithi ezininzi, okuvumela ukuba i-hardware ye-consumer-grade ifumaneka.

Kwimveliso ye-blockchain, sinikezela ama-miners abavela kwi-network problem solving kwi-two-time scales (i-short-to-medium-term) i-highest ability to select and order transactions. Ngokutsho, sinikezela into ye-”stake” kwi-Solana (i-money eyenza ukuba i-reward) kunye ne-proxy ye-hardware investment efana ne-Bitcoin, kodwa ngexesha efumaneka ukusetyenziswa kwe-Solana’s Tower Byzantine consensus.

Ngokufumaneka kwezi zophando zobuchwepheshe, sinika inkqubo apho ukhuseleko, i-decentralization, kunye ne-efficiency angakwazi ukuxhaswa kunokuba zihlanganisa - ukufumana i-0,1% verification overhead kunokuba i-10-1000x overhead yeengxaki ezahlukileyo.

Ishan Pandey:Inkqubo yakho inikeza ukunciphisa izindleko zokulungisa ngama-10x kunye ne-overhead ye-inference kwi-0.1%. Uyakwazi ukunciphisa iinkcukacha zobuchwepheshe ezinokufumaneka kwezi zixazululo?

Travis Good:Nceda nqakraza imibuzo yam ngexesha elandelayo malunga nenkqubo yethu yokufinyelela. Kwi-training, sisebenzisa izixhobo ezithile ezifanelekileyo zokufinyelela kunye nomnye ama-innovations ekusetyenziselwa ukusetyenziswa kwe-sparsity, oku kuxhomekeka ukuba iinkonzo phakathi kwezicwangciso zikhuselwe ngexesha lokufinyelela. Ukukhanyisa ngakumbi, nqakraza i-Litepaper yethu kwi-Ambient.xyz!

Ishan Pandey:Ukubalwa kwiminyaka emininzi ukuya kwiminyaka emininzi, iingcebiso ezininzi ze-technology okanye zebhizinisi zibonakalisa kwi-AI kunye ne-web3, kwaye njani i-Ambient ibekwe kwimveliso ezininzi?

Travis Good:Kwiintsuku ezidlulileyo ze-3 kuya kwiminyaka emininzi, ndingathanda izixazululo ezimbini ezininzi ezintsha i-AI kunye ne-web3:

Okokuqala, i-AI iya kubuyekeza ukusuka kubandakanyeka ngokubanzi kwi-human-directed ukuya kwi-autonomously. Siye kwakhona bakwazi ukuvelisa iinkqubo ze-agent that can plan, execute, and learn with minimal human supervision. Lezi zinkqubo ziya kubuyekeza i-demand for untrustworthy verification - akufunayo ukunxibelelana i-autority kwi-agent ye-AI unless unless unable to verify it is using the model and constraints you specified.

Okokuqala, iimfuno ze-computational yeemodeli ze-Frontier AI ziya kubandakanya ubunzima obuninzi, okwenza ukucindezeleka okuqinile. Iimodeli ezifanelekileyo ziya kufuneka izakhiwo ezininzi ezininzi ezininzi, okukhuthaza iingxaki malunga ne-access kunye ne-control. Thina ndingathanda ukubonisa oku kunye ne-Llama 4 ye-release.

Okwangoku, iimali ze-fiat ezivamile ziya kubandakanyeka kunye nokunyuka kwimveliso yobuchwepheshe kwiinkalo ze-digital apho abathengi be-AI zihlanganisa kunye nabanye abathengi be-AI, ngokuvamile kwiintlobo ze-jurisdiction. Oku kuya kubandakanyeka iimali ze-natively ye-digital kunye neengxaki ze-computing.

Okwangoku, iinkqubo zokulawula ziya ziyafumaneka kwi-AI, kunye neengxaki ye-transparency, i-auditability, kunye ne-safety. Iinkqubo ezinokukwazi ukunika iingcebiso ze-verifiable malunga nenkqubo zayo zokusebenza ziya kuthatha ukutya okwandisa.

I-Ambient ibekwe kwiindlela ezininzi zezi zixazululo:

Ukucinga
• inkqubo yethu yokuqinisekisa ebonakalayo xa inikeza ingqungquthela olufunekayo yeengxaki ze-AI ezisemthethweni.
Ukucinga
• Umgangatho wethu we-sharding model sinxulumene ngqo kwimeko ye-centralization.
Ukucinga
• Ngokusungula i-intelligence ye-machine njenge-standard ye-currency, sinikeza i-media ye-native ye-exchange ye-era elandelayo ye-agent-to-agent trade.
Ukucinga
• I-Architecture yethu yonke isakhiwo malunga ne-verification kunye ne-transparency.
Ukucinga

I-Vision ye-ultimate ye-create a layer of infrastructure that allows machine intelligence to flourish without becoming centralized under the control of a few corporations or governments.

Ngaba ushiye ukuthetha kwaye ushiye iinkcukacha!

I-Vested Interest Disclosure: Le umbhali uye umbhali wabanjwa ngokuhambelana nge-business blogging program yethu. I-HackerNoon uye uyifunyenwe kwi-report ngenxa yekhwalithi, kodwa iingcebiso ezininzi ziquka umbhali. #DYO